Final

Massachusetts Institute of Technology
6.042J/18.062J, Fall ’05: Mathematics for Computer Science December 21

Prof. Albert R. Meyer and Prof. Ronitt Rubinfeld revised December 22, 2005, 1118 minutes
Final Examination
Your name:
Circle the name of your Tutorial Instructor: David Hanson Jelani Sayan
This exam is closed book, but you may have a one page, twosided personal crib sheet.
There are 12 problems totaling 200 points. Write your solutions in the space provided, if
need be running over to the back of the page. Total time is 180 minutes. GOOD LUCK!
DO NOT WRITE BELOW THIS LINE
Problem Points Grade Grader
1 15
2 20
3 15
4 15
5 15
6 15
7 20
8 20
9 15
10 15
11 15
12 20
Total 200
Copyright © 2005, Prof. Albert R. Meyer and Prof. Ronitt Rubinfeld.

Final Examination Your name: 2
Problem 1 (15 points). Induction

Suppose S(n) is a predicate on natural numbers, n, and suppose
∀k ∈ N S(k) −→ S(k + 2). (1)
If (1) holds, some of the assertions below must always (A) hold, some can (C) hold but
not always, and some can never (N) hold. Indicate which case applies for each of the
assertions by circling the correct letter.
(a) (3 points) A N C (∀n ≤ 100 S(n)) ∧ (∀n > 100 ¬S(n))
(b) (3 points) A N C S(1) −→ ∀n S(2n + 1)
(c) (3 points) A N C [∃n S(2n)] −→ ∀n S(2n + 2)
(d) (3 points) A N C ∃n ∃m > n [S(2n) ∧ ¬S(2m)]
(e) (3 points) A N C [∃n S(n)] −→ ∀n ∃m > n S(m)

Problem 2 (20 points). State Machines

We will describe a process that operates on sequences of numbers. The process will start
with a sequence that is some permutation of the length 6n sequence
(1, 2, . . . , n, 1, 2, . . . , 2n, 1, 2, . . . , 3n).
(a) (5 points) Write a simple formula for the number of possible starting sequences.
If (s1 , . . . , sk ) is a sequence of numbers, then the i and jth elements of the sequence are out
of order if the number on the left is strictly larger the number on the right, that is, if i < j
and si > sj . Otherwise, the ith and jth elements are in order. Define p(S)::= the number
of “outoforder” pairs of elements in a sequence, S.
From the starting sequence, we carry out the following process:
(*) Pick two consecutive elements in the current sequence, say the ith and (i + 1)st.
I. If the elements are not in order, then switch them in the sequence and repeat step
(*).
II. If the elements are in order, remove both, resulting in a sequence that is shorter by
two. Then pick another element and remove it as well. If the length of the resulting
sequence is less than three, the process is over. Otherwise, reverse the sequence
and repeat step (*).
This process can be modelled as a state machine where the states are the sequences that
appear at step (*).
(b) (5 points) Describe a simple state invariant predicate that immediately implies that
if this process halts, then the final state is the sequence of length zero. (Just define the
invariant; you need not prove it has the requisite properties.)
(c) (10 points) Prove that this process always terminates by defining a nonnegative inte
ger valued derived variable that is strictly decreasing. (Just define the variable, you need
not prove it has these properties.)
Problem 3 (15 points). Equivalence Relations and Random Variables

A random variable, X is said to match a random variable, Y , iff X and Y are on the same
sample space and Pr {X �= Y } = 0. Prove that “matches” is an equivalence relation. Hint:
Note that Pr {X =
� Z} = Pr {[X �= Z] ∩ [X =� Y ]} + Pr {[X =
� Z] ∩ [X = Y ]}.
Problem 4 (15 points). Planarity.
(a) (8 points) Exhibit two planar drawings of the same 5vertex graph in which a face in
one drawing has more edges than any face in the other drawing.
(b) (7 points) Prove that all planar drawings of the same graph have the same number
of faces.
Problem 5 (15 points). Inclusionexclusion

A certain company wants to have security for their computer systems. So they have given
everyone a name and password. A length 10 word containing each of the characters:
a, d, e, f, i, l, o, p, r, s,
is called a cword. A password will be a cword which does not contain any of the subwords
”fails”, ”failed”, or ”drop”.
Use the Inclusionexclusion Principle to find a simple formula for the number of pass
words.
Problem 6 (15 points). Number Theory and Induction
(a) (5 points) Seashells are used for currency on a remote island. However, there are
only huge shells worth 210 dollars and gigantic shells worth 312 dollars. Suppose islander
A owes m > 0 dollars to islander B. Explain why the debt can be repaid through an
exchange of shells provided A and B both have enough of each kind.
(b) (10 points) Give an inductive proof that the Fibonacci numbers Fn and Fn+1 are rela
tively prime for all n ≥ 0. The Fibonacci numbers are defined as follows:
F0 = 0, F1 = 1, Fn = Fn−1 + Fn−2 (for n ≥ 2).

Problem 7 (20 points). Combinatorial Identities
(a) (10 points) Give a combinatorial proof that

n � �
� n
i = n2n−1 (2)
i=1
i
for all positive integers, n.
(b) (10 points) Now use the fact that the expected number of heads in n tosses of a fair
coin is n/2 to give a different proof of equation (2).
Problem 8 (20 points). Generating Functions

Let an be the number of ways to fill a box with n doughnuts subject to the following
constraints:
• The number of glazed doughnuts must be odd.
• The number of chocolate doughtnuts must be a multiple of 4.
• The number of plain doughnuts is 0 or 2.
• The number of sugar doughnuts is at most 1.
(a) (8 points) Write a generating function for each of the four doughnut types:
G(x) = C(x) =
P (x) = S(x) =
(b) (12 points) Derive a closed formula for an .

Problem 9 (15 points). Conditional Probability

There are 3 children of different ages. What is the probability that at least two are boys,
given that at least one of the two youngest children is a boy?
Assume that each child is equally likely to be a boy or a girl and that their genders are
mutually independent. A correct answer alone is sufficient. However, to be eligible for
partial credit, you must include a clearlylabeled tree diagram.
Problem 10 (15 points). Probability and Expectation

A box initially contains n balls, all colored black. A ball is drawn from the box at random.
• If the drawn ball is black, then a biased coin with probability, p > 0, of coming
up heads is flipped. If the coin comes up heads, a white ball is put into the box;
otherwise the black ball is returned to the box.
• If the drawn ball is white, then it is returned to the box.
This process is repeated until the box contains n white balls.

Let D be the number of balls drawn until the process ends with the box full of white balls.
Prove that E [D] = nHn /p, where Hn is the nth Harmonic number.
Hint: Let Di be the number of draws after the ith white ball until the draw when the
(i + 1)st white ball is put into the box.
Problem 11 (15 points). Deviation from the Mean

I have a randomized algorithm for calculating 6.042 grades that seems to have very
strange behavior. For example, if I run it more than once on the same data, it has dif
ferent running times. However, one thing I know for sure is that its expected running time
is 10 seconds.
(a) (5 points) What does Markov’s bound tell us about the probablity that my algorithm
takes longer than 1 minute (= 60 seconds)?
(b) (5 points) Suppose I decide to run the algorithm for 1 minute and if I don’t get an
answer by that time, I stop what I am doing, and completely restart from scratch. Each
time that I stop and restart the algorithm gives me an independent run of the algorithm.
So, what is an upper bound on the probability that my algorithm takes longer than 5
minutes to get an answer?
(c) (5 points) Suppose some 6.042 student tells me that they determined the variance of
the running time of my algorithm, and it is 25. What is an upper bound on the probability
that my algorithm takes longer than 1 minute?
Problem 12 (20 points). Estimation and Confidence

On December 20, 2005, the MIT fabrication facility produced a long run of silicon wafers.
To estimate the fraction, d, of defective wafers in this run, we will take a sample of n
independent random choices of wafers from the run, test them for defects, and estimate
that d is approximately the same as the fraction of defective wafers in the sample.
A calculation based on the Binomial Sampling Theorem (given below) will yield a near
minimal number, n0 , and such that with a sample of size n = n0 , the estimated fraction
will be within 0.006 of the actual fraction, d, with 97% confidence.
Theorem (Binomial Sampling). Let K1 , K2 , . . . , be a sequence of mutually independent 01

valued random variables with the same expectation, p, and let
n
�
Sn ::= Ki .
i=1
Then, for 1/2 > � > 0,
1 + 2�
2−n(1−H((1/2)−�))
�
� � �
� Sn �
Pr �
− p
�
≥ �
≤
� � ·
� (3)
n 2� 2π(1/4 − �2 )n
where
H(α) ::= −α log2 α − (1 − α) log2 (1 − α).
(a) (10 points) Explain how to use the Binomial Sampling Theorem to find n0 . You are
not expected to calculate any actual values, but be sure to indicate which values should
be plugged into what formulas.
(b) (10 points) The calculations in part (a) depend on some facts about the run and how
the n wafers in the sample are chosen. Write T or F next to each of the following state
ments to indicate whether it is True or False.
• The probability that the ninth wafer in the sample is defective is d.

• The probability that the ninth wafer in the run is defective is d.
• All wafers in the run are equally likely to be selected as the third wafer in the
sample.
• The expectation of the indicator variable for the last wafer in the sample being
defective is d.
• Given that the first wafer in the sample is defective, the probability that the
second wafer will also be defective is less than than d.
• Given that the last wafer in the run is defective, the probability that the next
tolast wafer in the run will also be defective is less than than d.
• It turns out that there are several different colors of wafer. Given that the first
two wafers in the sample are the same color, the probability that the first wafer is
defective may be < d.
• The probability that all the wafers in the sample will be different is nonzero.
• The probability that all choices of wafers in the sample will be different de
pends on the length of the run.
• The probability that all choices of wafers in the sample will be different is
−an
Θ(e ) for some constant a > 0.

Final

Uploaded by

Copyright:

Available Formats

Final

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Final

Uploaded by

Copyright:

Available Formats

Massachusetts Institute of Technology

6.042J/18.062J, Fall ’05: Mathematics for Computer Science December 21

Problem Points Grade Grader

Copyright © 2005, Prof. Albert R. Meyer and Prof. Ronitt Rubinfeld.

Problem 1 (15 points). Induction

∀k ∈ N S(k) −→ S(k + 2). (1)

(a) (3 points) A N C (∀n ≤ 100 S(n)) ∧ (∀n > 100 ¬S(n))

(b) (3 points) A N C S(1) −→ ∀n S(2n + 1)

(c) (3 points) A N C [∃n S(2n)] −→ ∀n S(2n + 2)

(d) (3 points) A N C ∃n ∃m > n [S(2n) ∧ ¬S(2m)]

(e) (3 points) A N C [∃n S(n)] −→ ∀n ∃m > n S(m)

Problem 2 (20 points). State Machines

(1, 2, . . . , n, 1, 2, . . . , 2n, 1, 2, . . . , 3n).

Problem 3 (15 points). Equivalence Relations and Random Variables

Problem 4 (15 points). Planarity.

Problem 5 (15 points). Inclusion­exclusion

Problem 6 (15 points). Number Theory and Induction

F0 = 0, F1 = 1, Fn = Fn−1 + Fn−2 (for n ≥ 2).

Problem 7 (20 points). Combinatorial Identities

(a) (10 points) Give a combinatorial proof that

for all positive integers, n.

Problem 8 (20 points). Generating Functions

• The number of glazed doughnuts must be odd.

• The number of chocolate doughtnuts must be a multiple of 4.

• The number of plain doughnuts is 0 or 2.

• The number of sugar doughnuts is at most 1.

(b) (12 points) Derive a closed formula for an .

Problem 9 (15 points). Conditional Probability

Problem 10 (15 points). Probability and Expectation

• If the drawn ball is white, then it is returned to the box.

This process is repeated until the box contains n white balls.

Problem 11 (15 points). Deviation from the Mean

Problem 12 (20 points). Estimation and Conﬁdence

Theorem (Binomial Sampling). Let K1 , K2 , . . . , be a sequence of mutually independent 0­1­

Then, for 1/2 > � > 0,

• The probability that the ninth wafer in the sample is defective is d.

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Problem 5 (15 points). Inclusionexclusion

Theorem (Binomial Sampling). Let K1 , K2 , . . . , be a sequence of mutually independent 01